A new index for scoring dental fluorosis, denoted by TSIF, recently developed at the NIDR, was evaluated for reproductibility by the two examiners who performed the fluorosis examinations. Each examiner was able to reproduce his fluorosis diagnosis upon repeated evaluation of a subsample of participants. The examiners experienced some difficulty in agreeing between themselves, especially in making the differentiation between the presence or absence of the very mild form of fluorosis. In spite of the difference between examiners, this index was able to detect statistically different levels of fluorosis between groups having optimal, 2, 3 and 4 times optimal levels of fluoride in their water supplies, after adjusting for examiner differences.